Every time you run a command you are evaling that code. And each eval has its own scope. Ie. my vars created within it disappear as soon as the eval ends.

But as you're just testing stuff, why worry about scoping?

In my repl, which is effectively similar to running in the debugger, I just forget about my (and our). And when I've tested the idea I just ^C & restart the repl to get a clean slate.

Definitely not recommended for real programs, but for just trying things it's fine. Less typing too :)